Wednesday, March 16, 2011
CMU Disaster Management Initiative Aids in Ongoing Relief Efforts
Carnegie Mellon Silicon Valley’s Disaster Management Initiative (DMI) has been working with various field agencies and first responders to aid relief efforts for the Japan earthquake and pacific tsunamis. The DMI has been working on multiple projects to assist on-the-ground, as well as with data, communication, and coordination.
CrisisCamp and Open Source Data
On Friday, March 11, Jeannie Stamberger, associate director of DMI, led CrisisCampSiliconValley, where volunteers arrived to develop country profiles and compiling data streams for the United Nations Office for the Coordination of Humanitarian Affairs (UN OCHA). Volunteers can continue to aid virtually by data mining and processing information at http://wiki.crisiscommons.org/wiki/Honshu_Quake. Ted Selker, associate director of the CyLab Mobility Research Center, was able to increase usability of the CrisisCommons Honshu wiki. “As the number of volunteers increases and the volume of information and sub-pages rapidly expands, the ability to see gather data quickly is essential,” said Dr. Stamberger.
Parsing Japanese Tweets
CMUSV students, faculty and volunteers have also been working on a project to parse Japanese tweets. The group is developing words and workflow and has begun to implement automated parsing of Japanese tweets for input into Google People finder, due to the high volume of Japanese tweets requesting help, medical care, and searching for loved ones.
Ian Lane, assistant research professor at CMUSV, whose expertise in machine language processing and twitter was essential to the project, and adjunct professor Jike Chong were able to work with the group to help accumulate keywords out of Japanese language Tweets including vocabulary to extract geo-location from the text and address technical difficulties such as format and characters in Japanese addresses. This project is now wrapping up, resulting in a tool for processing the initial surge of Tweets in any language, which the group continues to refine. “With Ian and Jike’s expertise, and the language expertise of several volunteers, students, and alumni, we were able to increase the amount of keywords in tweets, making categories for words such as “nuclear” or locations such as “Tokyo”, keeping in mind specific grammatical structure for given words,” explained Stamberger.
Assisting in Mapping
Trey Smith, systems scientist at CMUSV and member of the Intelligent Robotics Group at NASA Ames Research Center, was able to work with Urban Search and Rescue (USAR) teams to get accurate maps from Google physically into the hands of USAR-deployed teams to better assist those on the ground in the rescue efforts. “There is an urgent need for Japanese/English speakers to identify online local maps of areas in Japan by NetHope, an NGO deploying to the area,” said Stamberger. “CrisisCommons is the coordination group for that effort.”
Dr. Stamberger has also created a Google MyMap of first hand footage or reports of impact on the California coast, to illustrate the utility of crowdsourcing to the California Emergency Management Agency (CalEMA), which was activated for CrisisCampSiliconValley.
Jibbigo, a speech-to-speech translation application for iPhone and Android devices developed by researchers at Carnegie Mellon University, has already enjoyed popular demand for its Japanese-English version, listed as one of the top-selling iPhone apps in Japan. From now until March 31, 2011, Jibbigo is offering the Japanese-English speech translator at a 90% price reduction. The Jibbigo translator will function in areas that lack phone and internet connections. Jibbigo includes a robust medical vocabulary in both languages, which is ideal to assist the foreign aid worker. For more information, email firstname.lastname@example.org.
Ongoing Coordination among Groups
Carnegie Mellon’s DMI group is playing a role in coordinating communication between various groups. “Anyone can be a volunteer, and as relief efforts continue, it is good to get fresh sets of eyes on data. Those who can speak Japanese and English are especially helpful,” urges Stamberger.